On the Adversarial Robustness of Causal Algorithmic Recourse

Ricardo Dominguez-Olmedo, Amir H Karimi, Bernhard Schölkopf
Proceedings of the 39th International Conference on Machine Learning, PMLR 162:5324-5342, 2022.

Abstract

Algorithmic recourse seeks to provide actionable recommendations for individuals to overcome unfavorable classification outcomes from automated decision-making systems. Recourse recommendations should ideally be robust to reasonably small uncertainty in the features of the individual seeking recourse. In this work, we formulate the adversarially robust recourse problem and show that recourse methods that offer minimally costly recourse fail to be robust. We then present methods for generating adversarially robust recourse for linear and for differentiable classifiers. Finally, we show that regularizing the decision-making classifier to behave locally linearly and to rely more strongly on actionable features facilitates the existence of adversarially robust recourse.

Cite this Paper

BibTeX
@InProceedings{pmlr-v162-dominguez-olmedo22a, title = {On the Adversarial Robustness of Causal Algorithmic Recourse}, author = {Dominguez-Olmedo, Ricardo and Karimi, Amir H and Sch{\"o}lkopf, Bernhard}, booktitle = {Proceedings of the 39th International Conference on Machine Learning}, pages = {5324--5342}, year = {2022}, editor = {Chaudhuri, Kamalika and Jegelka, Stefanie and Song, Le and Szepesvari, Csaba and Niu, Gang and Sabato, Sivan}, volume = {162}, series = {Proceedings of Machine Learning Research}, month = {17--23 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v162/dominguez-olmedo22a/dominguez-olmedo22a.pdf}, url = {https://proceedings.mlr.press/v162/dominguez-olmedo22a.html}, abstract = {Algorithmic recourse seeks to provide actionable recommendations for individuals to overcome unfavorable classification outcomes from automated decision-making systems. Recourse recommendations should ideally be robust to reasonably small uncertainty in the features of the individual seeking recourse. In this work, we formulate the adversarially robust recourse problem and show that recourse methods that offer minimally costly recourse fail to be robust. We then present methods for generating adversarially robust recourse for linear and for differentiable classifiers. Finally, we show that regularizing the decision-making classifier to behave locally linearly and to rely more strongly on actionable features facilitates the existence of adversarially robust recourse.} }
Endnote
%0 Conference Paper %T On the Adversarial Robustness of Causal Algorithmic Recourse %A Ricardo Dominguez-Olmedo %A Amir H Karimi %A Bernhard Schölkopf %B Proceedings of the 39th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2022 %E Kamalika Chaudhuri %E Stefanie Jegelka %E Le Song %E Csaba Szepesvari %E Gang Niu %E Sivan Sabato %F pmlr-v162-dominguez-olmedo22a %I PMLR %P 5324--5342 %U https://proceedings.mlr.press/v162/dominguez-olmedo22a.html %V 162 %X Algorithmic recourse seeks to provide actionable recommendations for individuals to overcome unfavorable classification outcomes from automated decision-making systems. Recourse recommendations should ideally be robust to reasonably small uncertainty in the features of the individual seeking recourse. In this work, we formulate the adversarially robust recourse problem and show that recourse methods that offer minimally costly recourse fail to be robust. We then present methods for generating adversarially robust recourse for linear and for differentiable classifiers. Finally, we show that regularizing the decision-making classifier to behave locally linearly and to rely more strongly on actionable features facilitates the existence of adversarially robust recourse.
APA
Dominguez-Olmedo, R., Karimi, A.H. & Schölkopf, B.. (2022). On the Adversarial Robustness of Causal Algorithmic Recourse. Proceedings of the 39th International Conference on Machine Learning, in Proceedings of Machine Learning Research 162:5324-5342 Available from https://proceedings.mlr.press/v162/dominguez-olmedo22a.html.

Related Material